A single-institution experience with intraoperative in vivo confocal laser endomicroscopy for brain tumors in 50 patients

ObjectiveConfocal laser endomicroscopy (CLE), a handheld imaging technology, provides intraoperative real-time cellular resolution examination of tissue architecture. We evaluated the feasibility and diagnostic capability of the first clinically approved CLE system for intraoperative in vivo imaging of brain tumors.MethodsA total of 50 patients who were to undergo brain tumor surgery were prospectively enrolled. CLE images were interpreted by one CLE-experienced neuropathologist as lesional, non-lesional, or non-interpretable and compared to tissue histology acquired at the same location under neuronavigation. Diagnostic accuracy of CLE imaging was calculated using permanent sections as the standard for comparison. The neuropathologist provided real-time image interpretation using a built-in telepathology consultation platform in 27 cases.ResultsThe final pathology of the tumors in these patients included 28 gliomas, 5 meningiomas, 3 metastatic brain tumors, 5 treatment-related changes, and 10 other primary intracranial tumors. A total of 13,535 interpretable images were acquired from 304 regions of interest (ROIs). The first informative images were acquired within 10.5 s after the initiation of CLE imaging for each ROI. Mean CLE imaging time per case was 8.6 min. Using telepathology consultation extended CLE imaging per case time by 3.8 min (p=0.005). Communication between neurosurgeon and neuropathologist lasted 3.9 min per ROI. Overall sensitivity and specificity of CLE imaging were 93% and 81%, respectively. The specificity differed significantly between core and margin ROIs in glioma cases (93% vs. 50%, p=0.039). Diagnostic performance was not statistically different between new and recurrent glioma cases or between glioma and other tumor types.ConclusionsThe clinically approved CLE system allows intraoperative in vivo visualization of tissue histoarchitecture and identification of lesional tissue in real time, without the need for tissue biopsy and processing. CLE is efficient and cost-effective, with high diagnostic accuracy at the glioma core. However, CLE imaging at the tumor margin remains challenging.